Quickstart

Send your first request.

Create a Relay key, configure your client, and choose an available model from the console. These examples use the Chat Completions API.

Create API key

1Create a key
2Configure your client
3Send a model request

Base URL

https://proxystacks.com/v1

Model ID

YOUR_MODEL_ID

Replace YOUR_MODEL_ID with an available model ID from your console.

curl https://proxystacks.com/v1/chat/completions \
  -H "Authorization: Bearer sk-relay-your-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_MODEL_ID",
    "messages": [
      { "role": "user", "content": "Hello, Relay!" }
    ],
    "stream": true
  }'

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.RELAY_API_KEY,
  baseURL: "https://proxystacks.com/v1",
});

const stream = await client.chat.completions.create({
  model: "YOUR_MODEL_ID",
  messages: [{ role: "user", content: "Hello!" }],
  stream: true,
});

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["RELAY_API_KEY"],
    base_url="https://proxystacks.com/v1",
)

response = client.chat.completions.create(
    model="YOUR_MODEL_ID",
    messages=[{"role": "user", "content": "Hello!"}],
)

Authentication

Every model request must carry a Bearer token in the Authorization header. Relay stores only the SHA-256 hash of your API key.

Streaming

Set stream: true to get a standard SSE data stream. Balance deduction completes automatically once the stream ends.

Response format

Non-streaming responses use a standard JSON response format, plus a Relay request ID in the response header for easy log lookup.

AI coding tools

Point your AI coding tools at the Relay gateway. Copy a config block into the target file, then replace sk-relay-your-key and YOUR_MODEL_ID.

Merge into the env block of ~/.claude/settings.json, then restart Claude Code. ANTHROPIC_AUTH_TOKEN is sent as a Bearer token.

~/.claude/settings.json

{
  "env": {
    "ANTHROPIC_AUTH_TOKEN": "sk-relay-your-key",
    "ANTHROPIC_BASE_URL": "https://proxystacks.com/anthropic",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "YOUR_MODEL_ID",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "YOUR_MODEL_ID",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "YOUR_MODEL_ID",
    "ANTHROPIC_SMALL_FAST_MODEL": "YOUR_MODEL_ID"
  }
}

Save to ~/.codex/config.toml, then set the key in your shell: export RELAY_API_KEY=sk-relay-your-key. Codex uses the Responses API; Relay forwards statelessly, so keep disable_response_storage.

~/.codex/config.toml

model = "YOUR_MODEL_ID"
model_provider = "relay"
disable_response_storage = true

[model_providers.relay]
name = "Relay"
base_url = "https://proxystacks.com/v1"
env_key = "RELAY_API_KEY"
wire_api = "responses"

Save to ~/.config/opencode/opencode.json (or a project-level opencode.json), then pick model relay/YOUR_MODEL_ID.

~/.config/opencode/opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "relay": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Relay",
      "options": {
        "baseURL": "https://proxystacks.com/v1",
        "apiKey": "sk-relay-your-key"
      },
      "models": {
        "YOUR_MODEL_ID": {
          "name": "YOUR_MODEL_ID"
        }
      }
    }
  },
  "model": "relay/YOUR_MODEL_ID"
}

Run into trouble? [email protected]